Moldless electroplating for cylindrical microchannel fabrication

Transcription

1 Electrochemistry Communications 7 (2005) Moldless electroplating for cylindrical microchannel fabrication Yuheon Yi, Joo H. Kang, Je-Kyun Park * Department of BioSystems, Korea Advanced Institute of Science and Technology (KAIST), Guseong-dong, Yuseong-gu, Daejeon , Republic of Korea Received 23 June 2005; accepted 6 July 2005 Available online 8 August 2005 Abstract A conventional replication master made of photoresist for microchannel fabrication does not have reproducibility for the repeated replica molding, high temperature and high pressure processes. This paper presents how to fabricate cylindrical microfluidic channels easily, cheaply, and endurably by electroplating process. A hemispherical surface, instead of a rectangular surface, is achieved on an extremely thin metal seed layer by moldless electroplating. Without mold, exposed edges have abnormal growth rate and bad adhesion caused by high current density. However, with reduced thickness of the seed layer, the edge effect, converging electric field at the edge point, becomes negligible. A 5-lm wide gold strip was patterned on a glass wafer as seed layers. After the copper electroplating, a long and semicircular pole grew on the wafer. This copper protrusion played a role for the channel when poly(dimethylsiloxane) was poured. A depressed piece and a flat one built up a semicircular channel. Simulation results show that the tendency for metal to be semi-circle strongly depends on the aspect ratio of seed layers. Reversed connections to the power supply resolved copper into ions, which resulted in a dwindling of the radius of the copper pole when electroplating. Based on this fact, various diameters of channels were made by an electroplated replication master on a single wafer. Ó 2005 Elsevier B.V. All rights reserved. Keywords: Cylindrical microchannel; Electroplated copper master; Moldless electroplating; PDMS; Replication master 1. Introduction Since the concept of Ômicro total analysis system (ltas)õ was proposed [1], microfluidics which is performing precise controls of infinitesimal fluidic samples has been developed by silicon based microfabrication technologies [2,3] and thereafter, several techniques on the fabrication of micro/nanofluidic channels have been reported including silicon micromachining and polymer micropatterning [3 7]. Although silicon wet etching has been widely used to make cylindrical or semicircular microfluidic channels, it has some disadvantages as follows. Isotropically etched silicon chips are bonded with glass to give visual insights * Corresponding author. Tel.: ; fax: address: jekyun@kaist.ac.kr (J.-K. Park). for observation, or with symmetrically etched silicon to result in a complete circular cross section. From the above method, silicon wafer consumption is inevitable in the repeated microfluidic experiments. When there is a need to compare the effects of fluid characteristics caused by the different cross section areas of the channels, at least 4 wafers with different etched depths are required. Moreover, if an unwanted section area is made by an overetched wafer, the wafer cannot be used because the etching process is completely irreversible. The entire process for channel fabrication should start from the beginning. To achieve commercialization of the microfluidic chip technologies, polymer microfabrication method has been required rather than silicon or glass micromachining technique [8]. Fabrication strategy for a replication master is considered as one of the key issues in polymer microfabrication of the microfluidic channels, including /$ - see front matter Ó 2005 Elsevier B.V. All rights reserved. doi: /j.elecom

2 914 Y. Yi et al. / Electrochemistry Communications 7 (2005) deep reactive ion etching (DRIE), wet etching of silicon substrate [9], LIGA and electroplating with a photoresist (PR) mold [10], and SU-8 processing [6]. Among the polymer based fabrication processes, a rapid poly(dimethylsiloxane) (PDMS) casting technique with a PR replication master [6] has large influence upon the recent ltas or lab-on-a-chip development [11 13] because of its advantages: it is simple and easy to fabricate the channel patterns in short time and it also provides low-cost and reproducible processes. The PR replication masters, however, have some limitations. They cannot guarantee good durability and casting reliability because the repeated casting processes can cause the PR replication master worn out when the cured polymer is detached from the surface of replication master. Nor can they be used as a replication master for hot embossing and injection molding processes. Moreover, in some applications, the microchannel is required to have round cross-section [14], but the cross-section of the channel fabricated by the PR replication master is almost rectangular. In this study, it is shown that the above drawbacks can be avoided by moldless electroplating. A hemispherical surface, not a rectangular surface, is electroplated onto a seed layer without the PR mold. 2. Experimental Conventional micropatterning process was used to prepare the seed patterns for electroplating as follows. Table 1 Copper electroplating conditions Criteria Conditions Plating solution EEJA MicroFab Cu-300 Copper ingot Containing phosphorus 3% Current density ma cm 2 Temperature Room temperature Extraction rate 1.0 lm min 1 Chromium and gold were deposited on a Pyrex glass wafer (Corning, NY) up to 50 and 100 nm, respectively. AZ1512 (Clariant Corp., Somerville, NJ), positive PR was spin-coated on the wafer and the wafer was exposed to an ultraviolet (UV) light through a chrome mask. After developing the exposed wafer, the deposited gold and chrome except for the protected area by the PR were removed by a gold etchant (10 g of KI and 2.5 g of I 2 in 100 ml H 2 O) and a chrome etchant (CR-7SK; Cyantek Corp., Fremont, CA), respectively. Then, the PR stripping and cleaning processes in piranha solutions (a 3:1 mixture of concentrated H 2 SO 4 and 30% H 2 O 2 ) provided about 5 lm wide patterned seed layer for electroplating. The gold patterned wafer was soaked in copper electroplating solutions, Microfab Cu 300 (EEJA, Japan). The wafer was connected to minus terminal (traditionally black) of the DC power supply, Agilent E3646A (Agilent Technologies, CA) and the other positive terminal of the power supply was connected to a copper ingot containing 3% of phosphorous. The copper ingot was set inside of titanium basket. The other conditions are given in Table 1. Fig. 1 shows the scheme of electroplating and reverse electroplating by switching the power line connections. The microchannel was made by replica molding method [6]. The mixture (10:1) of prepolymer and curing agent (Sylgard 184; Dow Corning, Midland, MI) was poured on the completed copper protrusion. The mixture which was hardened in 80 C for 30 min was bonded with a slide glass for semicircular channel by air plasma treatment in 200 mtorr and 200 W for 10 s by an expanded plasma cleaner (PDC-002; Harrick Science, Ossing, NY). 3. Results and discussion An electrochemical deposition process of metal on cathode when a specific potential was applied between Fig. 1. The scheme of power line connection for copper (a) extraction or (b) dissolution on a wafer.

3 Y. Yi et al. / Electrochemistry Communications 7 (2005) anode and cathode is known as electroplating. Generally, the PR molds are usually needed for electroplating in order to obtain clearly defined structures. Because the electric fields tend to be uniformly distributed in media [15], a spherical surface, not a rectangular surface, can be electroplated onto a seed layer without mold. This process is called as Ômoldless electroplatingõ in this paper. It should be noted that electric fields have a tendency to collect at sharp edges of a conductor [15]. For a long time, electroplating technicians have been trying to design the target which has few sharp edges. Much higher current density at edges has resulted in rough surfaces and bad adhesion. To see the effect of non-uniformity of current at edges, it was required to check out the distribution of electric fields and equipotental lines by a computer simulation before the moldless electroplating. Fig. 2 Fig. 2. The simulation results generated by the CFD-ACE solver presenting cross-section of gold layer on a glass wafer. The parabolic lines are electric field intensity (V m 1 ). The growth profile of metal follows the lines. The ratios of the width to the height of seed layers are (a) 10:1, (b) 50:1, and (c) 100:1, respectively. shows the result of the simulation solved by the CFD-ACE solver (CFD-ACE; CFD Research Corporation, Huntsville, AL) with the condition of gold (resistivity, q, X m) pattern on a glass (q, X m) substrate surrounded by water (q, X m; permittivity, 81; permeability, 1) and constant current density of 30 ma cm 2. Grey scale is the intensity of electric fields. The darker grey scale shows the more intensified electric fields. The intensity of electric fields has different profiles according to the aspect ratio of the seed layers. The higher the seed layer, the stronger electric fields at edge. The profile will show a dumbbell shape, which has bulb like parts at the edge of the seed layer, when the aspect ratio of the seed layer is about 10:1 of the width to the height (Fig. 2(a)). Fig. 2(c), which has 100:1 aspect ratio, has proper distribution of electric fields that will form round cross section. To predict the final shape of the electroplated mold structures, the computational simulation for the electroplating process becomes significant. However, once the electroplating process begins, the different deposition rate at the initial phase is getting similar as the process is going on, because the electric flux tends to be uniformly distributed and the flux does not spontaneously go back to the non-uniform electric field distribution. Therefore, we have presented the electric field profile at the beginning part (t = 0 s) of the process to prove the reduced edge effect. An extremely thin seed layer can produce ideal cross section, but it was impossible to electroplate on a very thin seed layer because the thinner the seed, the higher the resistance. The thickness of tens of nanometer showed such high resistance that the seed could not endure the appropriate current density for bright electroplated surface. All the seeds were burn out and separated from the wafer during electroplating. Over one hundred nanometer thickness was the optimized point between the current density tolerance and the thickness maintaining 5 lm in width of the seed. The narrower width also resulted in higher resistance. About 50 lm high copper replication master was made through proper current and time during moldless electroplating process. The structures of the completed replication master were observed by optical microscope (Fig. 3(a) (c)) and the SEM (Fig. 3(d)). In order to reduce the height of the copper replication master structure, power supply terminals were swapped. In other words, the minus terminal was connected to the copper ingot and the other positive terminal to the wafer. After electroplating under reversed connection, the surface of the reduced copper replication master is not glistening because there is no reverse action of brightener in electrolyte solution (Fig. 4). However, if electroplating with normal power connection again, the surface becomes bright and glossy as the mold grows. Therefore, the

4 916 Y. Yi et al. / Electrochemistry Communications 7 (2005) Fig. 3. With metal seed pattern of 5 lm in width (a), electroplating is performed to have about 30 lm in height of replication master after 30 min plating (b), and about 50 lm in height after 55 min (c), and (d) the SEM image of cross-section of 50 lm-replication master structure. Fig. 4. About 25 lm in height of copper replication master structure obtained from the 50 lm in height electroplated structure after 10 min from the reversed power connection. The surface is not so much bright as the pictures in Fig. 3(b). Fig. 5. The SEM image of the cross-sectional view of PDMS channel after curing on a copper replication master. The ripples on the surface of the channel result from the shrinkage of PDMS. The channel width is about 40 lm. channel size can be controlled by either time control under given growth rate or by alternating power connection from the normal to the reverse. Fig. 5 is the SEM image showing the PDMS channel before bonding. It seems that the ripples on the surface of the channel result from the shrinkage of PDMS. This results show the disadvantages of this moldless electroplating method, such as shapes, profiles, and processing problems due to the moldless option. Also this method is not adequate for the structure of high aspect ratio, because the height and width of the channel are simultaneously increasing when it is electroplated. However, the fabrication technique for metal replication master is adaptable to the multilayer soft lithography [14] because of its round cross-section of the channel. In addition, if the symmetric PDMS pieces for circular channel are aligned, then a round channel could be formed.

5 Y. Yi et al. / Electrochemistry Communications 7 (2005) Conclusion From the experiments, a durable metal replication master for cylindrical microchannel was fabricated through moldless electroplating. The dimension of the master was controlled by the electroplating time. Swapping power line connections during electroplating resulted in the reduction of the cross-section, which can also control the dimension of the master. Consequently, various diameters of channels were made. It was also confirmed that copper masters with curved surfaces for subsequent PDMS molding was achieved by electroplating onto a thin and long metal seed without photoresist mold. Further research is underway to integrate the microfluidic device based on the moldless electroplating. Acknowledgement This research was supported by a Grant (04K ) from Center for Nanoscale Mechatronics & Manufacturing, one of the 21st Century Frontier Research Programs, Ministry of Science and Technology, Korea. References [1] A. Manz, N. Graber, H.M. Widmer, Sens. Actuators B Chem. 1 (1990) 244. [2] Y. Ning, G. Fitzpatrick, Microfabrication Processes for Silicon and Glass Chips, in: Biochip Technology, Harwood Academic Publishers, New Jersey, 2001, p. 17. [3] A. Manz, J.C. Fettinger, E. Verpoorte, H. Lüdi, H.M. Widmer, D.J. Harrison, Trends Anal. Chem. 10 (1991) 144. [4] A. Rasmussen, M. Gaitan, L.E. Locascio, M.E. Zaghloul, J. Microelectromech. Syst. 10 (2001) 286. [5] L. Martynova, L.E. Locascio, M. Gaitan, G.W. Kramer, R.G. Christensen, W.A. MacCrehan, Anal. Chem. 69 (1997) [6] D.C. Duffy, J.C. McDonald, O.J.A. Schueller, G.M. Whitesides, Anal. Chem. 70 (1998) [7] N.R. Tas, J.W. Berenschot, P. Mela, H.V. Jansen, M. Elwenspoek, A. van den Berg, Nano Letters 2 (2002) [8] H. Becker, C. Gärtner, Electrophoresis 21 (2000) 12. [9] M.B. Esch, S. Kapur, G. Irizarry, V. Genova, Lab Chip 3 (2003) 121. [10] H. Becker, U. Heim, Sens. Actuators A Phys. 83 (2000) 130. [11] D.R. Reyes, D. Iossifidis, P.-A. Auroux, A. Manz, Anal. Chem. 74 (2002) [12] P.-A. Auroux, D. Iossifidis, D.R. Reyes, A. Manz, Anal. Chem. 74 (2002) [13] T. Vilkner, D. Janasek, A. Manz, Anal. Chem. 76 (2004) [14] M.A. Unger, H.-P. Chou, T. Thorsen, A. Scherer, S.R. Quake, Science 288 (2000) 113. [15] P.M. Fishbane, S. Gasiorowicz, S.T. Thornton, Physics for Scientists and Engineers, Prentice-Hall, New Jersey, 1996, pp